This is a discussion on using sparse random network models as prior distributions in graphical models.